How does this differ from standard plan execution?

Unlike parallel session execution, this skill stays in your current session and dispatches a fresh subagent for every single task to prevent context pollution and accumulated errors.

Can I use this for tightly coupled tasks?

It is best used for tasks that are mostly independent. For extremely coupled changes where logic overlaps significantly, manual execution may be more effective.

Is code review mandatory between tasks?

Yes, the skill is designed to enforce a code review after each task is completed to catch critical or important issues before the next task begins.

What other skills are required to use this effectively?

This skill integrates with 'writing-plans' for the initial roadmap, 'requesting-code-review' for quality gates, and 'finishing-a-development-branch' for final delivery.

Subagent-Driven Development

Name: Subagent-Driven Development
Author: Krosebrook

byKrosebrook

•

Productivity & Workflow

Streamlines implementation by dispatching fresh subagents for individual tasks with automated quality gates and code reviews.

Subagent-Driven Development is a specialized workflow skill for Claude Code designed to execute complex implementation plans with high precision and minimal context pollution. By breaking down plans into independent tasks and assigning each to a fresh subagent, the skill ensures that development remains focused and isolated. Between every task, an automated code review acts as a quality gate, catching issues early and maintaining architectural integrity. This workflow is ideal for developers who want to stay within a single session while achieving the rigor of a multi-step development process, combining the speed of AI automation with the reliability of continuous testing and peer-style review.

Key Features

01Integrated Test-Driven Development (TDD) enforcement for subagents

02Real-time implementation tracking using TodoWrite task management

031 GitHub stars

04Context isolation via fresh subagent dispatch for every task

05Automated quality gates with mandatory code reviews between steps

06Seamless handoff to branch verification and final merge workflows

Use Cases

01Executing a multi-feature development plan within a single Claude Code session

02Refactoring large codebases where independent modules require isolated attention

03Accelerating iteration cycles while maintaining strict code quality and testing standards

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add krosebrook/source-of-truth-monorepo subagent-driven-development

For use in Claude.ai and ChatGPT

Download Skill